Computational Thermochemistry - American Chemical Society

Af H°(Cx Hy Fz , 0K) = x Af H°(Cgas , 0K) + y Af H°(Hgas , 0 K) + z Af H°(Fgas , 0 K). -ID 0 ( C x H y F z ). (2). Finally, the room temperature e...
1 downloads 0 Views 2MB Size
Chapter 18

Ab Initio Calculations for Kinetic Modeling of Halocarbons Downloaded by STANFORD UNIV GREEN LIBR on September 28, 2012 | http://pubs.acs.org Publication Date: February 1, 1998 | doi: 10.1021/bk-1998-0677.ch018

1

1,2

R. J. Berry , M. Schwartz , and Paul

1,2

Marshall

1

Center for Computational Modeling of Nonstructural Materials, Wright Laboratory, Materials Directorate, Wright-Patterson A i r Force Base, O H 45433 Department of Chemistry, University of North Texas, Denton, TX 76203 1

The thermochemistry and reaction kinetics of halogenated hydrocarbons have been investigated by ab initio methods in order to improve our understanding of their flame chemistry and likely roles in flame suppression. Bond additivity corrections at the G2, G2(MP2), CBS-4 and CBS-Q levels of theory were developed for fluorinated and chlorinated C and C species, including saturated and unsaturated compounds. The resulting enthalpies of formation are in excellent agreement with experimental values. Transition states for the reactions of H atoms with hydrofluoromethanes were characterized at up to the G2 level of theory, and application of transition state theory yielded rate constants in good accord with experimental results. A similar analysis for H and O H reactions with CH I also agrees with the known thermochemistry and kinetics. These investigations provide insight into the major product channels and the temperature dependence of the rate constants. The implications for flame suppression by haloalkanes are discussed. 1

2

3

During the past two decades, it has become apparent that the release of volatile chlorofluorocarbons (CFCs) and halon fire suppressants (e.g. CF Br, CF ClBr, 3

2

C F B r - C F B r ) is a major cause of depletion of the stratosphere's ozone layer (7-5). Hence, there have been numerous restrictions placed upon the industrial use of CFCs and halons, which has led to concerted efforts to find new, "ozone-friendly" replacements (4). Potential replacement agents include hydrofluorocarbons and iodocarbons, which are degraded in the troposphere and, hence, pose no significant risk to the ozone layer. They also possess low global warming potentials. The effectiveness of a proposed flame suppressant can be reliably predicted via kinetic modeling provided accurate thermochemical and kinetic data are available for the hundreds of reactions associated with the suppressant and its interactions with 2

2

© 1998 American Chemical Society

In Computational Thermochemistry; Irikura, K., et al.; ACS Symposium Series; American Chemical Society: Washington, DC, 1905.

341

COMPUTATIONAL THERMOCHEMISTRY

342

Downloaded by STANFORD UNIV GREEN LIBR on September 28, 2012 | http://pubs.acs.org Publication Date: February 1, 1998 | doi: 10.1021/bk-1998-0677.ch018

flame species. Such data are often difficult to obtain from experiments, especially at combustion temperatures. Experimental determinations are further hampered by the transient nature of many of the intermediate species involved, which makes it difficult to isolate individual reactions and species for measurements. An ab initio computational alternative is desirable because it would enhance our understanding of these complex processes at a molecular level and provide a cost-effective way of screening/predicting suitable replacement agents for fire suppression. Thermochemistry of Haloalkanes Enthalpies of Formation from ab Initio Energies. Ab initio quantum mechanics has proven to be a most valuable tool for estimating thermochemical quantities (dissociation energies, enthalpies of formation, heat capacities, entropies, etc.) of gas phase molecules, radicals and ions in systems where experimental data are either unavailable or unreliable. Briefly outlined below is the procedure by which one may utilize quantum mechanical energies to furnish a direct estimate of the enthalpy of formation of a fluoroalkane. The molecular atomization energy, Z D , is obtained from the calculated 0

energies of the molecule and constituent atoms via the relation: L D ( C H F ) = x E(calc, C) + y E(calc, H) + z E(calc, F) - E (calc, C H F ) 0

x

y

z

0

x

y

z

(1)

E for the molecule contains the zero-point energy (ZPE), obtained from calculated vibrational frequencies after adjustment by the appropriate scale factor (0.8929 for HF/6-3 lG(d) frequencies). The enthalpy of formation at 0 K can then be computed from the atomization energy and the experimental enthalpies of formation of the constituent atoms via: 0

A H°(C H F , 0K) = x A H°(C f

x

y

z

f

gas

, 0K) + y A H°(H f

gas

, 0 K) + z A H ° ( F f

gas

, 0 K)

-ID (C H F ) 0

x

y

(2)

z

Finally, the room temperature enthalpy is obtained from: A H°(298.15 K) = A H°(0 K) + 8 H ° ( C H F ) -x 8H°(C) - 0.5y 8 H ° ( H ) - 0.5z 8H°(F ) f

f

x

y

z

2

2

(3)

8H° = H°(298.15 K) - H°(0 K) represents the thermal contribution to a species' enthalpy. It is obtained from experimental data for the elements. For the compound, one computes this quantity from the calculated rotational constants and scaled vibrational frequencies using standard statistical mechanical formulae. Current state of the art ab initio methods such as Pople's G2 [Gaussian-2] (5, see chapter by Curtiss and Raghavachari) and simpler G2(MP2) (6) methods, and Petersson's CBS-4 and CBS-Q methods (7, see chapter by Petersson), have been proposed for the accurate determination of enthalpies of formation using the procedure

In Computational Thermochemistry; Irikura, K., et al.; ACS Symposium Series; American Chemical Society: Washington, DC, 1905.

18.

BERRY ET A L .

343

Kinetic Modeling of Halocarbons

outlined above. The estimated accuracies of these methods, as ascertained by comparison with the "G2 test set" (5) of 55 molecules with accurately known atomization energies, are very good. For the test set, RMS deviations of the computed 1

energies from experiment for the four protocols are 6.3, 6.7, 9.2, and 4.2 kJ mol" , respectively (7). The agreement is not surprising since the four methods were optimized to minimize disagreement with experimental data including the "G2 test set." However, these data include only one molecule with a carbon-halogen bond, Downloaded by STANFORD UNIV GREEN LIBR on September 28, 2012 | http://pubs.acs.org Publication Date: February 1, 1998 | doi: 10.1021/bk-1998-0677.ch018

CH3CI. Therefore, a systematic investigation of the ability of these methods to predict accurate enthalpies of formation of chlorofluorocarbons was conducted. Chlorofluoromethanes. A previous study by Ignacio and Schlegel (8) utilized HF/6-3 lG(d) optimized geometries and vibrational frequencies, MP4/6-31G(d,p) energies and isodesmic reactions (using the experimental enthalpies of C H , C F and CCI4) to compute the enthalpies of formation of the remaining chlorofluoromethanes. 4

4

1

Based on the results they estimated the accuracy of this method to be ±13 kJ mol" . In order to test the capabilities of the currently utilized high level "compound methods" to calculate the enthalpies of halocarbons with improved accuracy, the G2(5), G2(MP2) (6), CBS-Q and CBS-4 (7) methods were applied (9) to the chlorofluoromethanes

which

included

the

four

fluoromethanes,

the

four

chloromethanes and the six compounds containing both chlorine and fluorine. In I the calculated G2 and CBS-Q enthalpies are compared with experiment. Experimental enthalpies of the fluoromethanes were taken from Kolesov's compilation (10). For the remaining molecules the recommended J A N A F (77) enthalpies were chosen for comparison.

Deviations

of

the

calculated

enthalpies

from

experiment,

|A H°(calc) - A H°(expt)], are listed in parentheses. An alternative, although less f

f

complete, compilation of enthalpies of formation of the CFCs has been published by 1

Pedley et al. (12). The literature values from this reference are within 2 kJ mol" of those in the J A N A F tables (77) for most species. However, for C F C 1 and CFC1 , the 2

2

3

1

enthalpies in the former reference are lower by 14 and 20 kJ mol" , respectively. The computed enthalpies in I exhibit substantial deviations from experiment, 1

with RMS errors of 15 and 19 kJ mol" for G2 and CBS-Q, respectively. These deviations are systematic, with almost all calculated enthalpies lying lower than reported experimental values, as evidenced by the large negative mean deviations. These results are in sharp contrast to many earlier investigations of non-halogenated organic species; e.g. RMS deviations from experimental enthalpies for the "G2 test 1

set" were typically 4-9 kJ mol" for these methods (vide supra). To explore the distribution of errors in these series in greater detail, it is instructive to plot the deviation from experiment, [A H°(calc) - A H°(expt)], as a f

f

function of one type of carbon-halogen bond while holding the number of the other C-X bond types constant, e.g. a plot of error versus n

C F

(number of C F bonds) in the

series CH C1, CH FC1, CHF C1, CF C1. This plot is displayed for the CBS-Q method 3

2

2

3

in 1 (one finds similar trends using the other methods). One observes quite clearly that

In Computational Thermochemistry; Irikura, K., et al.; ACS Symposium Series; American Chemical Society: Washington, DC, 1905.

COMPUTATIONAL THERMOCHEMISTRY

344

Downloaded by STANFORD UNIV GREEN LIBR on September 28, 2012 | http://pubs.acs.org Publication Date: February 1, 1998 | doi: 10.1021/bk-1998-0677.ch018

the negative error increases monotonically with increasing number of either C-F or CCl bonds; the trend is "roughly" linear with C-X bond. One approach to correct systematic errors in ab initio estimates of enthalpies of formation is to employ the concept of Bond Additivity Corrections (B ACs), developed by Melius and coworkers (13,14, see chapter by Zachariah and Melius) for MP4/ 6-31G(d,p) enthalpies. In this method, it is assumed that the deviation of calculated enthalpies from experiment is a linear function of the number of each type of bond in the molecule, as indicated in the following expression: A H°(BAC) f

= A H°(calc) - Z n A f

{

{

= A H°(calc) - [ncH A c + n f

H

C F

AQ + n c F

C1

ACQ]

(4)

Table I. Calculated and Experimental Enthalpies of Formation in Chlorofluoro­ methanes Species

3

Expt.

G2

b

CBS-Q

b

G2

b

CBS-Q

b

[BAC]

[BAC] -74.9±0.4

-77.7(-2.8)

-74.0(0.9)

-77.7(-2.8)

-74.0(0.9)

-232.6±8.4

-244.1 (-11.5)

-238.7(-6.1)

-237.6(-5.0)

-235.2(-2.6)

-452.2±1.8

-463.7(-11.5)

-457.6(-5.4)

-450.7(1.5)

-450.6(1.6)

-697.6±2.7

-714.0(-16.4)

-706.7(-9.1)

-694.5(3.1)

-696.2(1.4)

-933.0±1.7

-956.5(-23.5)

-947.7(-14.7)

-930.5(2.5)

-933.7(-0.7)

-83.7±2.1

-85.5(-1.8)

-86.3(-2.6)

-82.7(1.0)

-77.8(5.9)

-95.5±1.3

-98.1 (-2.6)

-105.6(-10.1)

-92.4(3.1)

-88.6(6.9)

-103.2+1.3

-107.6(-4.4)

-125.3(-22.1)

-99.2(4.0)

-99.8(3.4)

-96.0±2.1

-107.7(-11.7)

-137.3(-41.3)

-96.5(-0.5)

-103.3(-7.3)

CH FC1

-261.9±13.0

-273.3(-11.4)

-272.3(-10.4)

-264.0(-2.1)

-260.3(1.6)

CHF C1

-481.6±13.0

-498.1 (-16.5)

-495.9(-14.3)

-482.3(-0.7)

-480.4(1.2)

-707.9±3.3

-731.8(-23.9)

-728.2(-20.3)

-709.4(-1.5)

-709.2(-1.3)

-283.3±13.0

-295.8(-12.5)

-301.4(-18.1)

-283.7(-0.4)'

-280.9(2.4)

-491.6±8.0

-513.9(-22.3)

-517.4(-25.8)

-495.3(-3.7)

-493.4(-1.8)

-288.7±6.3

-305.2(-16.5)

-318.8(-30.1)

-290.3(-1.6)

-289.8(-l.l)

CH

4

CH F 3

CH F 2

CHF CF

2

3

4

CH C1 3

CH C1 2

2

CHCI3 CCI4 2

2

CF3CI CHFC1 CF C1 2

CFCI3

2

2

RMS

±6.9

14.5

18.8

2.6

3.4

AVG



-12.6

-15.3

-0.2

0.7

a

1

A H ° at 298.15 K in kJ moF units Values in parentheses represent deviations from experiment SOURCE: Reproduced from reference 9. Copyright 1996 American Chemical Society. f

In Computational Thermochemistry; Irikura, K., et al.; ACS Symposium Series; American Chemical Society: Washington, DC, 1905.

18.

BERRY E T A L .

345

Kinetic Modeling of Halocarbons

Linear regression was used to fit equation 4 to the experimental data on the series of 15 CFCs. This provided values for the B A C corrected enthalpies of formation in I and the three B A C parameters, AQ , AQ and A C Q (II (A)) for all four ab initio H

methods. The values of A ^

H

F

were quite small, and were therefore set equal to zero.

One sees from I that the B A C corrected enthalpies of formation are in extremely good 1

agreement with experiment. The residual RMS deviations of 2.6 and 3.4 kJ mol" , are almost an order of magnitude lower than errors in the uncorrected enthalpies and, Downloaded by STANFORD UNIV GREEN LIBR on September 28, 2012 | http://pubs.acs.org Publication Date: February 1, 1998 | doi: 10.1021/bk-1998-0677.ch018

1

indeed, lie significantly below the RMS experimental uncertainty of 6.9 kJ mol" . The BACs have also removed the systematic underprediction of the enthalpies of

Table II. Fluoroalkane Bond Additivity Parameters and Resulting Errors Method

ACF

RMSl

(A) Chlorofluoromethanes (A^p + A c )

b

AVEl

b

RMS2

C

AVE2

5

C

d

C 1

G2

-6.51±0.41

0.0

14.5

-12.6

2.6

-0.2

G2(MP2)

-7.98±0.38

0.0

21.5

CBS-4

-1.28±0.74

0.0

20.7

-19.3

2.4

0.0

-15.3

4.7

0.6

CBS-Q

-3.51±0.55

0.0

18.8

-15.3

3.4

0.7

(B) Fluoroethanes (A^p) G2

-6.51±0.41

0.0

28.4

-25.8

7.6

-6.3

G2(MP2)

-7.98±0.38

0.0

33.4

-29.7

7.3

-5.8

CBS-4

-1.28±0.74

0.0

10.9

-10.0

8.6

-6.5

CBS-Q

-3.51±0.55

0.0

20.9

-18.4

9.7

-7.7

(C) Fluoroethanes (A(-p + f ) c

G2

-6.51±0.41

1.24±0.08

28.4

-25.8

5.3

-1.6

G2(MP2)

-7.98±0.38

1.20±0.06

33.4

-29.7

4.9

-1.0

CBS-4

-1.28±0.74

1.73+1.50

10.9

-10.0

7.2

-2.6

CBS-Q

-3.51±0.55

1.78±0.29

20.9

-18.4

5.3

-0.8

(D) Fluoroethylenes G2

-6.51±0.41

1.24±0.08

21.4

-18.2

4.7

-2.1

G2(MP2)

-7.98±0.38

1.20±0.06

25.2

-21.0 '

4.9

-1.9

CBS-4

-1.28±0.74

1.73±1.50

10.5

-8.7

5.8

-3.8

CBS-Q

-3.51±0.55

1.78±0.29

17.3

-13.7

5.2

-2.0

a

In units of kJ mol Errors in ab initio enthalpies Errors in B A C corrected enthalpies Using a A C Q o f - 2 . 8 0 ± 0 . 4 1 , -6.54±0.38, -10.62±0.74 and -8.50±0.55, respectively, for the G2, G2(MP2), CBS-4 and CBS-Q methods SOURCE: Reproduced with permission from reference 19. Copyright 1997 Elsevier. b

c

d

In Computational Thermochemistry; Irikura, K., et al.; ACS Symposium Series; American Chemical Society: Washington, DC, 1905.

COMPUTATIONAL THERMOCHEMISTRY

Downloaded by STANFORD UNIV GREEN LIBR on September 28, 2012 | http://pubs.acs.org Publication Date: February 1, 1998 | doi: 10.1021/bk-1998-0677.ch018

346

0

1

2

3

4

0

1

No. of C-F Bonds

2

3

4

No. of C - C l Bonds

Figure 1. Deviations of CBS-Q enthalpies of formation of chlorofluoromethanes from experiment plotted as a function of the number of: (A) C-F bonds and (B) C - C l bonds (Reproduced from reference 9. Copyright 1996 American Chemical Society.)

formation, as revealed by the extremely small average errors in the corrected results. Hence, B A C corrections for the G2 and CBS-Q methods were successfully employed to predict highly accurate enthalpies of formation in the chlorofluoromethanes. As shown by the statistical data presented in II (A), similar improvements were also obtained with the G2(MP2) and CBS-4 methods (9). The curvature exhibited in 1(B) and the increasingly negative slopes of the lines in 1(A) are manifestations of "heavy atom" interactions between the C-F and C-Cl bonds (9). The application of spin-orbit coupling corrections to atomic energies can improve the agreement of G2 enthalpies of formation with experiment. These corrections (9)

can remove some,

but not all of the systematic

errors in

chloromethanes, and are far too small to account for the observed deviations in fluoromethanes and the mixed species. Therefore, spin-orbit corrections were not performed explicitly, but are absorbed by the empirical B A C parameters. Fluoroethanes and Ethylenes. Next, the G2, G2(MP2), CBS-Q and CBS-4 quantum mechanical protocols were utilized to compute the enthalpies of formation of C fluorocarbons. Contained in III (A) is a comparison of the computed G2 enthalpies 2

for fluoroethanes with experiment (15-18). The deviations in the ab initio enthalpies are also presented in the Table and plotted (for the G2 and CBS-Q methods only) in 2 (closed squares and solid lines).

In Computational Thermochemistry; Irikura, K., et al.; ACS Symposium Series; American Chemical Society: Washington, DC, 1905.

18.

BERRY E T A L .

347

Kinetic Modeling of Halocarbons

One observes from the RMS and average errors in the table that the G2 enthalpies exhibit large negative deviations from experiment. Furthermore, from 2, one sees that these negative errors are systematic with an approximately linear dependence upon the number of C-F bonds in the molecule.

Table III. C Fluorocarbons: Calculated vs. Experimental Enthalpies

3

Downloaded by STANFORD UNIV GREEN LIBR on September 28, 2012 | http://pubs.acs.org Publication Date: February 1, 1998 | doi: 10.1021/bk-1998-0677.ch018

2

Species

Expt.

G2

b

G2[BAC]

b

(A) Fluoroethanes CH3-CH3

-84.1 ±0.4

-86.0 (-1.9)

-86.0 (-1.9)

-263.2±1.6

-279.7 (-16.5)

-271.6 (-8.4)

2

-433.9±11.8

-459.8 (-25.9)

-443.7 (-9.8)

2

-500.8±6.3

-516.4 (-15.6)

-500.3 (0.5)

-664.8±4.2

-687.0 (-22.2)

-662.8 (2.0)

-745.6± 1.6

-772.1 (-26.5)

-747.9 (-2.3)

-877.8±17.6

-906.6 (-28.8)

-874.3 (3.5)

CH -CH F 3

2

CH F-CH F 2

CH -CHF 3

CH F-CHF 2

2

CH3-CF3

CHF -CHF 2

2

CH F-CF

3

-895.8±4.2

-934.2 (-38.4)

-901.9 (-6.1)

CHF -CF

3

-1104.6±4.6

-1145.9 (-41.3)

-1105.5 (-0.9)

-1342.7±6.3

-1383.7 (-41.0)

-1335.3 (7.4)

2

2

CF3-CF3

RMS

±7.7

28.4

5.3

AVG



-25.8

-1.6

(B) Fluoroethylenes CH =CH

52.4±0.8

53.3 (0.9)

53.3 (0.9)

-140.1±2.5

-146.2 (-6.1)

-138.1 (2.0)

CHF=CHF[Z]

-297.1±10.0

-315.1 (-18.0)

-299.0 (-1.9)

CHF=CHF[E]

-292.9±10.0

-318.4 (-25.5)

-302.3 (-9.4)

-336.4±4.0

-359.2 (-22.8)

-343.1 (-6.7)

-491.0±9.0

-512.5 (-21.5)

-488.3 (2.7)

-658.5±2.9

-693.2 (-34.7)

-660.9 (-2.4)

2

2

CH =CHF 2

CH =CF 2

2

CHF=CF CF =CF 2

2

2

RMS

±6.7

21.4

4.7

AVG



-18.2

-2.1

a

-1

A H° in units of kJmol Values in parentheses represent deviations from experiment SOURCE: Reproduced with permission from reference 19. Copyright 1997 Elsevier. f

In Computational Thermochemistry; Irikura, K., et al.; ACS Symposium Series; American Chemical Society: Washington, DC, 1905.

COMPUTATIONAL THERMOCHEMISTRY

348

(B) CBS-Q

(A) G2

fl

o o o

Downloaded by STANFORD UNIV GREEN LIBR on September 28, 2012 | http://pubs.acs.org Publication Date: February 1, 1998 | doi: 10.1021/bk-1998-0677.ch018

O"" "



uncorrected

o

B A C corrected

_ J

0

2 4 No. of C-F Bonds

0

6

,

L

,

o

I

2 4 No. of C-F Bonds

6

Figure 2. Deviations of the computed enthalpies of formation of fluoroethanes from experiment as a function of the number of C-F bonds. (A) G2 and (B) CBS-Q (Reproduced with permission from reference 19. Copyright 1997 Elsevier.)

Since the deviations are linearly dependent upon n , one may, in principle, C F

again apply the B A C correction to obtain corrected results. Furthermore, because of the close similarity of the fluoroethanes (C s) to the fluoromethanes (Qs) in the 2

earlier work, it would be reasonable to expect that the same C - F B A C parameter should also correct the systematic errors found in this study. The results of this procedure are displayed in category B of II. The second column contains the values of A

C F

(9) with their estimated standard deviations. The last two pairs of columns

represent the RMS and average errors without and with the application of the C-F B A C parameter. As noted above, the values of RMS2 and AVE2 in II (A) demonstrate that the application of C-F and C-Cl BACs removes virtually all systematic errors in the chlorofluoromethanes. From RMS2 in II (B), one sees, of course, that transferring the C-F B A C to fluoroethanes yields a substantial decrease in RMS deviations in the C s. However, in contrast to the Cjs, the comparatively large negative values of AVE2 2

reveal that not all of the negative systematic error has been removed by the bond additivity correction. While the problem could be remedied by refitting to minimize the RMS residuals in the fluoroethanes, this yields a larger value for A p which would C

overcorrect enthalpies in the Q s . Further, it is physically unrealistic that the error due to a C-F bond should differ in the two series. In order to explore the source of the remaining systematic errors in the fluoroethanes, results of the earlier investigation of the chlorofluoromethanes (9) were

In Computational Thermochemistry; Irikura, K., et al.; ACS Symposium Series; American Chemical Society: Washington, DC, 1905.

18.

BERRY E T A L .

349

Kinetic Modeling of Halocarbons

re-examined. In that work, it was discussed at some length how trends and curvature in plots of A H°(calc) - A H°(expt) vs. n . (at fixed n . ) and vs. n . f

f

c

F

c

c l

c

c l

(at fixed n . ) c

F

provided definitive evidence of "heavy atom" interactions. In the earlier work, it was decided not to include heavy atom interactions since the RMS residuals using linearly independent BACs

were already below

the experimental uncertainties in the

chlorofluoromethanes. In contrast, for the fluoroethanes, the comparatively large residual errors (RMS2 in II (B)) and negative average deviations (AVE2) indicate that

Downloaded by STANFORD UNIV GREEN LIBR on September 28, 2012 | http://pubs.acs.org Publication Date: February 1, 1998 | doi: 10.1021/bk-1998-0677.ch018

the introduction of a heavy atom interaction parameter is necessary in this series to account for increased errors due to the presence of a second carbon atom attached to the carbon containing the C - C bond. Thus an interaction parameter, f (79), was c

incorporated into the B A C equation which, assuming no C - H bond error, becomes: A H°(BAC) = A H°(calc) - n f

f

C F

A ^f

(5)

c

The value of the parameter is optimized (holding A^p constant at the value determined for the CFCs) to minimize the RMS deviation from experiment in the fluoroethanes. The resultant values of the interaction parameter with its error estimate are shown in the third column of II (C). RMS2 of category C shows that the residuals using a C-F B A C with an interaction parameter are reduced significantly (relative to category B) for all but the CBS-4 method. Further, the much smaller negative values of AVE2, relative to category B, indicate that almost all of the remaining systematic error has been removed from the B A C corrected enthalpies of formation. Values of RMS2 are well within the RMS experimental uncertainty (III (A)). The removal of systematic error is also demonstrated in 2, in which it is seen that errors in the B A C corrected enthalpies of formation (open circles and dashed line) are clustered about A H°(calc) f

A H°(expt) = 0. f

These studies have been extended to the fluoroethylenes ( C 2 H F . , x=0-4) X

4

X

and fluoroacetylenes ( C H F . , x=0-2) using the G2, G2(MP2), CBS-Q and CBS-4 2

X

2

X

methods (20). It was decided to use the same parameter values for A^p and f as were c

used for the fluoroethanes. As shown in III (B) for the G2 method, systematic errors in the calculated enthalpies (third column) are almost completely removed by application of the B A C and interaction parameter (fourth column). It is very satisfying to find that, for

both methods,

the RMS deviations

in B A C corrected enthalpies

of the

fluoroethylenes are lower than RMS uncertainties in the experimental data (11(D)). Extension of ab Initio Methods to Heavier Halogens. The G2 method was originally defined for the elements H through CI. However, Radom and coworkers have recently developed basis sets for Br and I that permit G2 analysis of species containing these halogens (27). These authors described both effective core potential and all-electron basis sets that lead to G2[ECP] and G2[AE] energies, respectively. The G2[ECP] method is much faster computationally, as only the valence electrons are treated explicitly, and there is claimed to be little loss of accuracy as compared to G2[AE] results

(27).

These basis sets have been employed to analyze the

In Computational Thermochemistry; Irikura, K., et al.; ACS Symposium Series; American Chemical Society: Washington, DC, 1905.

COMPUTATIONAL THERMOCHEMISTRY

350

thermochemistry of bromine and iodine monoxides (22) and hydroxides (23,24) via G2 and related approaches. Hassanzadeh and Irikura have analyzed similar systems using CCSD(T) calculations with large basis sets (25), and Lee has investigated a series of triatomic bromine compounds at a similar level of theory (26). An early application of high-level methods to Br and I-containing compounds was by Kello and Sadlej, who studied cyanogen halides (27). Typically, a variety of isodesmic or homodesmic reactions have been employed to relate the unknown heats of formation Downloaded by STANFORD UNIV GREEN LIBR on September 28, 2012 | http://pubs.acs.org Publication Date: February 1, 1998 | doi: 10.1021/bk-1998-0677.ch018

to experimental quantities. A complication is that radicals with significant unpaired electron spin density close to Br or I nuclei exhibit significant spin-orbit coupling, a relativistic effect which is excluded from standard ab initio calculations, but which can lead to large energy corrections. This effect leads to splitting of electron levels. For example, standard calculations on ground-state halogen atoms yield the weighted mean energy of the two 2

Pj levels (J = 1/2 and 3/2), weighted by the 2J+1 degeneracy. However, at 0 K only the 2

lower energy P

3 / 2

level is populated. This correction is usually neglected for F and CI 2

atoms, because the P j levels lie close together and the errors introduced are only 1.6 1

and 3.5 kJ mol" , respectively. These corrections are implicit in the C - F and C-Cl BACs discussed here, and have recently been shown to improve agreement with experimental energies of Cl-containing compounds (28). By contrast, for I atoms at 0 1

K the required correction is 30.3 kJ mol" and must be included if chemical accuracy is to be achieved. The magnitude of the spin-orbit splitting between ground-state sublevels for radicals may be computed (21) or, if available, empirical data can be employed. On the assumption that the spin-orbit states have similar geometries and vibrational frequencies, the influence of low-lying electronic states on thermodynamic functions such as C , S and H - H is straightforward to evaluate (77), once the p

T

0

energy splittings and degeneracies are established. Haloalkane Kinetics Product Channels for H + Fluoromethanes. The decomposition of fluoromethanes in hydrocarbon flames has recently been investigated (29) using numerical simulations in conjunction with flame speed measurements. In stoichiometric methane/air flames at high (8%) inhibitor concentrations, the relative amounts of fluoromethane destroyed via reaction with H atoms, unimolecular decomposition and reaction with O H radicals were found to be 4:3:3 and 6:3:1, respectively, for C H F and C H F . 2

2

3

Destruction by H attack, which is the leading fluoromethane consumption pathway, has been the subject of several experimental and theoretical investigations. H + CH _ F 4

X

X

-> products

x=l,2,3,4

(6-9)

There are three possible reaction channels; abstraction of F, abstraction of H , and substitution. For C H F all three channels are thermochemically reasonable (77): 3

In Computational Thermochemistry; Irikura, K., et al.; ACS Symposium Series; American Chemical Society: Washington, DC, 1905.

18.

BERRY E T A L .

H + CH F 3

-> C H + HF 3

->CH F + H 2

2

2 9 8

= -111 kJ mol"

A H°

2 9 8

=-16 kJmol"

A H°

2 9 8

=-12 k J m o r

r

4

1

A H° r

->CH + F

r

Thus, the net decomposition rate, k$ = k^ + k a

Downloaded by STANFORD UNIV GREEN LIBR on September 28, 2012 | http://pubs.acs.org Publication Date: February 1, 1998 | doi: 10.1021/bk-1998-0677.ch018

351

Kinetic Modeling of Halocarbons

1

1

(6a) (6b) (6c)

+ k$ . In their review Baulch et

6 b

c

al. (30) suggested F-abstraction is the dominant channel. Their rate constant recommendation is largely based on the work of Westenberg and deHaas (31) who monitored the disappearance of C H F in the presence of a large excess of H, so that 3

knowledge of the products was not necessary for the determination of the total rate constant. Similarly, Parsamyan and Nalbandyan (32) assumed F-abstraction to be the main pathway while analyzing their H + C H F data. On the other hand BAC-MP4 2

2

calculations of Westmoreland et al. (33) predict H-abstraction to be the most important pathway for both reactions 6 and 7. Recently, high level ah initio methods such as G2(MP2) and G2(ZPE=MP2) have been employed to compute the reaction barriers for these pathways; full details may be found elsewhere (34). H-abstraction is computed to have the lowest energy barrier, E * , for reactions 6-8. For reaction 9 the barrier to F-substitution is very high 0

which leaves F-abstraction as the only accessible channel. Controversy also exists over the value of the rate constants. For example, measurements of k^ differ by over two orders of magnitude at 600 K, where different measurement techniques overlap (31,35). To resolve these discrepancies the G2(MP2) results (34) were used in transition state theory (TST) calculations to predict the reaction rate constants, k

T S T

, as a function of temperature. The resulting Arrhenius

plots of the rate constant are compared with experiment in 3.

_ TST "

Q

k T B

K

1

h Q

H

Q

E*'

% TS*

"RT C

C

H

X

P

(10)

F

4- x x where T is the quantum mechanical tunneling correction derived at the zero-curvature level (36) and the partition functions, Q, include rotational symmetry numbers. This is the simplest form of TST. Q was derived on the usual basis of the rigid rotor harmonic oscillator approximation. Any anharmonicity would lead to slightly higher values of Q, with a likely larger effect on the TS because of its looser modes. Thus, k may be underestimated. However, an opposing factor is that neglect of variational effects implies k will tend to be too high. The zero-curvature tunneling model neglects two opposing effects. One is that this one-dimensional model does not allow for multi-dimensional corner-cutting tunneling paths, which would increase T. On the other hand, the MP2 imaginary frequencies employed here are probably too high, which will lead to overestimated T values. A new method for improved calculations of T S T

T S T

In Computational Thermochemistry; Irikura, K., et al.; ACS Symposium Series; American Chemical Society: Washington, DC, 1905.

352

COMPUTATIONAL THERMOCHEMISTRY

the energy barrier is outlined in Petersson's chapter in this book. The effects of these neglected factors appear either to be small or to cancel because, as shown below, the canonical TST model generally agrees well with experiment. For H + C H F , the minor F-abstraction and substitution channels are predicted 3

to have small rate constants, hence the total rate constant k is essentially equal to k^. 6

This agrees with the recommended rate constant in the 600-1000 K range within the

Downloaded by STANFORD UNIV GREEN LIBR on September 28, 2012 | http://pubs.acs.org Publication Date: February 1, 1998 | doi: 10.1021/bk-1998-0677.ch018

factor of 5 experimental error limits (30). If the entire difference between theory and experiment is assigned to errors in the ab initio E * for this reaction, then this 0

1

corresponds to a computational error of only 4 kJ mol" (at 800 K). Employing the critical evaluation by Baulch et al. (30), some of the earlier measurements are seen to be in error (37,38). The computed rate constant for reaction 7 was within a factor of 2 of the value reported by Parsamyan and Nalbandyan (32). For reaction 8 the computed rate constant compares well with the reported experimental determinations (39-41). For reaction 9 (H + C F ) , the calculated rate constant is of the same order of magnitude as 4

the measurements by Kochubei and Moin (42). However, the measurements imply that 1

the computed G2(MP2) barrier has been overestimated by about 22 kJ mol" . Thus the ab initio potential energy surfaces suggest that, contrary to some earlier assumptions, H atoms react predominantly with the C - H bonds in fluoromethanes and that the dominant product is H . F atom abstraction is unfavorable 2

*

ref. (31)

• r e f . (32) • r e f . (35) o

ref. (37) ref. (38)

• • - ref. (39) ref. (40) *

ref. (41)

* - • ref. (42)

0

1 2 3 4 1000 K/Temperature

0

1 2 3 4 1000 K/Temperature

Figure 3. H + fluoromethanes: G2(MP2) Arrhenius plots (solid lines) vs. experiment (Reproduced with permission from reference 34. Copyright 1997 Elsevier.)

In Computational Thermochemistry; Irikura, K., et al.; ACS Symposium Series; American Chemical Society: Washington, DC, 1905.

18.

BERRY ET A L .

Kinetic Modeling of Halocarbons

353

kinetically, even though HF formation is the most exothermic pathway. G2(MP2) based TST results are in good accord with experimentally determined rate constants for H attack on C H F , C H F and C H F . C F is several orders of magnitude less reactive than the other fluoromethanes towards atomic hydrogen, because it does not contain labile C - H bonds. The high C-F bond strength also makes unimolecular dissociation unfavorable under combustion conditions. Slow F-atom abstraction is the only plausible pathway for H-atom attack on C F . Thus C F is essentially inert in a flame and the flame suppressant activity of C F is largely physical, i.e. through cooling and dilution. By contrast, the other fluoromethanes react quickly with H atoms to yield F-containing radicals that undergo further chemistry. This allows for the possibility of chemical flame suppression radical scavenging by C H F , C H F and CHF . 3

2

2

3

4

4

4

Downloaded by STANFORD UNIV GREEN LIBR on September 28, 2012 | http://pubs.acs.org Publication Date: February 1, 1998 | doi: 10.1021/bk-1998-0677.ch018

4

3

2

2

3

The Reactions of H Atoms and O H Radicals with Iodomethane. As a demonstration of the application of G2 analysis to iodine chemistry, results obtained via the G2[AE] method of Radom and coworkers (27) are presented for the reactions of H and O H with CH I. These data were used to assess the combustion chemistry of 3

CH I, 3

an important intermediate in the chemistry of iodine-mediated flame

suppression by agents such as C F I (43). First the thermochemistry was examined, as a 3

check of the accuracy of the G2 method, and then the reaction kinetics were analyzed and compared with experiment. One aim was to distinguish between various possible sets of products, which for some systems has been a difficult experimental problem. The present work describes an ah initio analysis of the kinetics of the reactions CH I + H

->CH I + H

3

2

(11a)

2

- > C H + HI

(lib)

3

CH I + OH 3

-> C H I + H 0

(12a)

- > C H + HOI

(12b)

2

2

3

The results are employed to derive high-temperature rate constants via T S T and to assess branching ratios for H vs I abstraction. Contributions from the I-displacement channels were assumed to be negligible based on the G2(AE) results reported for CH I 3

+ H -> C H + I (44). 4

The transition state for reaction 11a was first investigated by Schiesser et al. using moderately sized ECP basis sets (45), while details of the G2 analysis of channels 11 a-12b may be found elsewhere (46). As a check on the accuracy of the G2 methodology, A H ° r

2 9 8

was computed for the C - H and C-I bond breaking reactions 11a 1

and l i b . The results are -9.0 and -61.7 kJ m o l , in excellent accord with the 1

experimental values of -5.3 ± 6.7 and -62.2 ± 0.9 kJ mol" (47), and thus the thermochemistry of the C - H and C-I bonds is seen to be well-described at the G2 level.

In Computational Thermochemistry; Irikura, K., et al.; ACS Symposium Series; American Chemical Society: Washington, DC, 1905.

COMPUTATIONAL THERMOCHEMISTRY

354

As in the H + fluoromethane example discussed above, the rate constants k were compute^ via canonical TST (see equation 10). The scaled HF/6-3 lG(d) and MP2=full/6-31G(d) frequencies of the TSs are similar, except for the low frequency C-I-H bending mode. This apparently reflects the neglect or overestimation of the influence of electron correlation on this mode, which has a significant influence on the TST rate constant. As a compromise, the geometric mean of TST calculations based

Downloaded by STANFORD UNIV GREEN LIBR on September 28, 2012 | http://pubs.acs.org Publication Date: February 1, 1998 | doi: 10.1021/bk-1998-0677.ch018

on both sets of frequencies was derived. The rate constants are plotted in 4. The results for H + C H I are consistent with the known bond strengths D 3

weaker C-I bond ( D (D

2 9 8

(47), where the

1

2 9 8

= 239 kJ mol" ) is much more reactive than the C - H bond

1

2 9 8

= 431 kJ mol" ). This is in accord with previous experimental studies which

concluded or assumed that HI was the dominant product (48-51), and yields a k

n

which agrees well with the room temperature experimental data. TS structures for attack by the O H radical at the H and I atoms of C H I were also investigated. In the latter case no barrier beyond the endothermicity was discernible on the HF/6-3 lG(d) potential energy surface. Consideration of the QCISD(T)/6-311G(d,p) energies calculated at selected points along the MP2/6-31G(d) intrinsic reaction coordinate showed that the energy is below that of HOI + C H at all points. A distinct TS therefore cannot be localized and canonical T S T cannot be applied to this channel. k was roughly estimated via the relation k = A exp(-E /RT) 3

3

1 2 b

:

~ ~

Iff

1

\

(A)

-

=

ffl

r

+ CHj



\ \

\

R

\

1 IOT" - H^oyV o o

• \

f 10"

r

O y +Hp

\

1

iff

3

(B)

\

o

S

a

\ 1 0

,

1

,

1

2

i\

ref. (48)

CHJ+HOI

'.

.

r e f ( 4 9 )

°

ref. (50) '.

0

ref. (57)

0

ref. (57) ~



ref. (55)

,

r

1

1

4

3

"

'•

0

,

1

1

,

1 2

,

1 3

1000 K/Temperature

1000 K/Temperature

Figure 4. Computed TST results (solid lines) vs. experiment for: (A) H + CH I. (B) O H + CH3I. The dashed line for the C H + HOI channel shows an empirical estimate. (Reproduced with permission from reference 46. Copyright 1997 Elsevier.) 3

3

In Computational Thermochemistry; Irikura, K., et al.; ACS Symposium Series; American Chemical Society: Washington, DC, 1905.

18.

BERRY E T A L .

355

Kinetic Modeling of Halocarbons

by assuming a pre-exponential factor A = 1 x 10

3

cm molecule

s , equal to that 1

measured for C F I + OH (52), and an activation energy E = 23 kJ mol" . This is equal 3

a

1

to the endothermicity at 298 K based on D ( H O - I ) = 216 kJ mol" (52). 298

4(B) shows that, by contrast to H atom attack, the bond that is more reactive towards O H is the stronger C - H bond and that C H I + H 0 formation, the more 2

2

exothermic channel, is expected to dominate at all temperatures. A check on the Downloaded by STANFORD UNIV GREEN LIBR on September 28, 2012 | http://pubs.acs.org Publication Date: February 1, 1998 | doi: 10.1021/bk-1998-0677.ch018

reliability of the kinetic calculations is comparison with the measurements of the total removal rate constant k

1 2

=k

1 2 a

+k

1 2 b

made by Brown et al. (53). They are seen to be

in excellent agreement, as is the room temperature k

1 2

value obtained by Gilles et

al. (51). Again, the geometric mean of the HF and MP2-based TST results gives good accord with experiment and provides a TST extrapolation of these measurements to combustion conditions. The k

1 2

value from the extrapolation to 2000 K is about 25

times greater than the value obtained from a simple linear Arrhenius extrapolation of the data of Brown et al. (55), which reflects the significant curvature of the theoretical Arrhenius plot. This predicted curvature is similar to that measured for the analogous reaction of O H with C H (54). The TST analysis also indicates that HOI formation by 4

OH attack on C H I is of only minor importance in both flames and the atmosphere. 3

The TST rate constants for reactions 11 and 12 and experimental data (51,55) for O + CH I

-> IO and other products

(13)

C H I + Ar

- > C H + I + Ar

(14)

3

3

3

were employed in a numerical simulation of an adiabatic premixed stoichiometric CH /air flame at atmospheric pressure, modeled with C H E M K I N (56) using the GRI4

Mech mechanism (57). The results apply to trace quantities of C H I (i.e. the C/H/O 3

chemistry was assumed to be unaffected). As may be seen from 5, the dominant removal pathway is via H-atom attack. The next most important removal pathway, about an order of magnitude less effective than reaction 11, is attack by OH. Thus C H I will be a minor product of C H I decay, with HI as the dominant product. 2

3

Summary Ab initio methods have been applied to investigate the thermochemistry of halogenated alkanes and the kinetics of major pathways for their reactions in flames. Where the computed thermochemistry and rate expressions can be compared with experimental values, the results are in good accord and provide a means to extrapolate limited temperature information over wider temperature ranges. The ab initio calculations of enthalpies of formation and rate constants are sufficiently accurate to survey a wide range of possible steps in combustion mechanisms, in order to help limit the number of processes to be included in combustion models and to aid in the identification of key reactions that merit more detailed laboratory study.

In Computational Thermochemistry; Irikura, K., et al.; ACS Symposium Series; American Chemical Society: Washington, DC, 1905.

COMPUTATIONAL THERMOCHEMISTRY

356

2000

10° 10' tJ

6

IA

10'

L

O H attack

H attack

H

1500 H

n>

Downloaded by STANFORD UNIV GREEN LIBR on September 28, 2012 | http://pubs.acs.org Publication Date: February 1, 1998 | doi: 10.1021/bk-1998-0677.ch018

l(f

13

1000

O attack 4

2 io > o

3

io ,

/ /

,

500

/ 2

io

3 n> •-t 2 eo S 3

unimolecular decay

1/

10' 0.00

0.02

0.04

0.06

0.08

0.10

Distance from burner (cm) Figure 5. Reciprocal lifetimes for traces of C H I 3

in an adiabatic premixed

stoichiometric CH /air flame at atmospheric pressure with respect to unimolecular 4

decomposition and attack by H , O and O H (solid lines, left axis), and the temperature profile (dashed line, right axis). (Reproduced with permission from reference 46. Copyright 1997 Elsevier.)

Acknowledgments The authors thank the Air Force Office of Scientific Research and the Wright Laboratory, Materials Directorate at Wright-Patterson Air Force Base. M.S. and P.M. acknowledge the Robert A. Welch Foundation (Grant Nos. B-657 and B-1174) and the UNT Faculty Research Fund for financial support. Literature Cited 1. 2. 3. 4.

Rowland, F. S. Environ. Sci. Technol. 1991, 25, 622. McFarland, M.; Kaye, J. Photochem. Photobiol. 1992, 55, 911. Banks, R. E. J. Fluorine Chem. 1994, 67, 193. (a) Todd, C. S. The Use of Halons in the United Kingdom and the Scope for Substitution; HMSO: London, 1991, and references therein, (b) Grosshandler, W. L.; Gann, R. G.; Pitts, W. M. Evaluation of Alternative In-Flight Fire Suppressants for Full-Scale Testing in Simulated Aircraft Engine Nacelles and Dry Bays; NIST Spec. Publ. 861; NIST: Washington DC, 1994. 5. Curtiss, L. A.; Raghavachari, K.; Trucks, G. W; Pople, J. A. J. Chem. Phys. 1991, 94, 7221.

In Computational Thermochemistry; Irikura, K., et al.; ACS Symposium Series; American Chemical Society: Washington, DC, 1905.

Downloaded by STANFORD UNIV GREEN LIBR on September 28, 2012 | http://pubs.acs.org Publication Date: February 1, 1998 | doi: 10.1021/bk-1998-0677.ch018

18.

BERRY E T A L .

Kinetic Modeling of Halocarbons

357

6. Curtiss, L. A.; Raghavachari, K.; Pople, J. A. J. Chem. Phys. 1993, 98, 1293. 7. Ochterski, J. W.; Petersson, G. A.; Montgomery, J. A., Jr. J. Chem. Phys. 1996, 104, 2598. 8. Ignacio, E. W.; Schlegel, H. B. J. Phys. Chem. 1992, 96, 5830. 9. Berry, R. J.; Burgess, D. R., Jr.; Nyden, M. R.; Zachariah, M. R.; Melius, C. F.; Schwartz, M. J. Phys. Chem. 1996, 100, 7405. 10. Kolesov, V. P.; Russ. Chem. Rev. 1978, 47, 1145. 11. Chase, M. W., Jr.; Davies, C. A.; Downey, J. R., Jr.; Frurip, D. J.; McDonald, R. A.; Syverud, A. N. "JANAF Thermochemical Tables, Third Edition". J. Phys. Chem. Ref. Data, 1985, 14, Suppl. 1. 12. Pedley, J. B.; Naylor, R. D.; Kirby, S. P. Thermochemical Data of Organic Compounds; Chapman & Hall: London, UK, 1986. 13. (a) Melius, C. F. Springer-Verlag DFVLR Lecture Notes; Springer-Verlag: Berlin, 1990. (b) Melius, C. F. Chemistry and Physics of EnergeticMaterials;S. N. Kluwer Academic: New York, 1992. 14. (a) Ho, P.; Melius, C. F. J. Phys. Chem. 1990, 94, 5120. (b) Allendorf, M. D.; Melius, C. F. J. Phys. Chem. 1993, 97, 72. 15. Tsang, W.; Hampson, R. F. J. Phys. Chem. Ref. Data 1986, 15, 1087. 16. Chen, S. S.; Rodgers, A. S.; Chao, J.; Wilhoit, R. C.; Zwolinski, B. J. J. Phys. Chem. Ref. Data 1975, 4, 441. 17. Burgess, D. R., Jr.; Zachariah, M. R.; Tsang, W.; Westmoreland, P. R. Prog. Energy Combust. Sci. 1996, 21, 1034. 18. Lacher, J. A.; Skinner, H. A. J. Chem. Soc. (A) 1968, 1034. 19. Berry, R. J.; Ehlers, C. J.; Burgess, D. R., Jr.; Zachariah, M. R.; Nyden, M. R.; Schwartz, M. J. Mol. Struct. (Theochem), in press. 20. Berry, R. J.; Schwartz, M. Struct. Chem., submitted. 21. Glukhovtsev, M. N.; Pross, A.; McGrath, M. P.; Radom, L. J. Chem. Phys. 1995, 103, 1878. 22. McGrath, M. P.; Rowland, F S. J. Phys. Chem. 1996, 100, 4815. 23. McGrath, M. P.; Rowland, F. S. J. Phys. Chem. 1996, 98, 4774. 24. Glukhovtsev, M. N.; Pross, A.; Radom, L. J. Phys. Chem. 1996, 100, 3498. 25. Hassanzadeh, P.; Irikura, K. K. J. Phys. Chem. 1997, 101, 1580. 26. Lee, T. J. J. Phys. Chem. 1995, 99, 15074. 27. Kellö, V; Sadlej, A. Mol. Phys. 1992, 75, 209. 28. Curtiss, L. A.; Raghavachari, K.; Redfern, P. C.; Pople, J. A. J. Chem. Phys. 1997, 106, 1063. 29. Linteris, G. T. In Halon Replacements: Technology and Science; Miziolek, A. W.; Tsang, W., Eds.; ACS Symp. Ser. 611; ACS: Washington DC, 1995; p 260. 30. Baulch, D. L.; Duxbury, J.; Grant, S. J.; Montague, D. C. J. Phys. Chem. Ref. Data 1981, 10, suppl. 1. 31. Westenberg, A. A.; deHaas, N. J. J. Chem. Phys. 1975, 62, 3321. 32. Parsamyan, N. I.; Nalbandyan, A. B. Arm. Khim. Zhur. 1968, 21, 1003. 33. Westmoreland, P. R.; Burgess, D. F., Jr.; Tsang, W.; Zachariah, M. R. 25th Symp. (Int.) Combust.; The Combustion Institute: Pittsburg, PA, 1994; p 1505. 34. Berry, R. J.; Ehlers, C. J.; Burgess, D. F., Jr.; Zachariah, M. R.; Marshall, P. Chem. Phys. Lett., 1997, 269, 107. 35. Aders, W. -K.; Pangritz, D.; Wagner, H. Gg. Ber. Bunsenges. Phys. Chem. 1975, 79, 90. In Computational Thermochemistry; Irikura, K., et al.; ACS Symposium Series; American Chemical Society: Washington, DC, 1905.

COMPUTATIONAL THERMOCHEMISTRY

Downloaded by STANFORD UNIV GREEN LIBR on September 28, 2012 | http://pubs.acs.org Publication Date: February 1, 1998 | doi: 10.1021/bk-1998-0677.ch018

358

36. Steckler, R.; Hu,W.-P.;Liu,Y.-P.;Lynch, G. C.; Garrett, B. C.; Isaacson, A. D.; Lu, D.-H.; Melissas, V. S.; Truong, T. N.; Rai, S. N.; Hancock, G. C.; Lauderdale, J. G.; Joseph,T.;Truhlar, D. G. POLYRATE version 6.5 (University of Minnesota, Minneapolis, 1995). 37. Hart, L. W.; Grunfelder, C.; Fristrom, R. M. Combust. Flame 1974, 23, 109. 38. Jones, W. E.; Ma, J. L. Can. J. Chem. 1986, 64, 2192. 39. Amphlett, J. C.; Whittle, E. Trans. Faraday Soc. 1967, 63, 2695. 40. Skinner, G. B.; Ringrose, G. H. J. Chem. Phys. 1965, 43, 4129. 41. Richter, H.; Vandooren, J.; Van Tiggelen, P. J. 25th Symp. (Int.)Combust.;The Combustion Institute: Pittsburg, PA, 1994; p 825. 42. Kochubei,V.F.;Moin, F. B. Kinet. Catal. 1971, 11, 86. English translation p. 712. 43. Noto, T.; Babushok, V.; Burgess, D. R., Jr.; Hamins, A.; Tsang, W.; Miziolek, A. 26th Symp. (Int.) Combust.; The Combustion Institute: Pittsburg, PA, in press. 44. Yuan, J.; Wells, L; Marshall, P. J. Phys. Chem. A, in press. 45. Schiesser, C. H.; Smart, B. A.; Tran, T.-A. Tetrahedron 1995, 51, 3327. 46. Marshall, P.; Misra, A.; Berry, R. J. Chem. Phys. Lett. 1997, 265, 48. 47. McMillen, D. F.; Golden, D. M. Ann. Rev. Phys. Chem. 1982, 33, 493. Data extrapolated to 0 K using ab initio H - H values. 48. Leipunskii, I. O.; Morozov,I.I.;Tal'roze, V. L. Dokl. Phys. Chem. 1971, 198, 547. Russ. orig. p. 136. 49. Levy, M. R.; Simons, J. P. J. Chem. Soc. Faraday Trans. 2 1975, 71, 561. 50. Sillesen, A.; Ratajczak, E.; Pagsberg, P. Chem. Phys. Lett. 1993, 201, 171. 51. Gilles, M. K.; Turnipseed, A. A.; Talukdar, R. K.; Rudich, Y; Villalta, P. W.; Huey, L. G.; Burkholder, J. B.; Ravishankara, A. R. J. Phys. Chem. 1996, 100, 14005. 52. Berry, R. J.; Yuan, W.-J.; Misra, A.; Marshall, P. unpublished work. 53. Brown, A.C.;Canosa-Mas, C. E.; Wayne, R. P. Atmos. Environ. 1990, 24, 361. 54. Madronich, S.; Felder, W. 20th Symp. (Int.) Combust.; The Combustion Institute: Pittsburg, PA, 1984; p 703. 55. Saito, K.; Tahara, H.; Kondo, O.; Yokubo, T; Higashihara, T; Murakami, I. Bull. Chem. Soc. Jpn. 1980, 53, 1335. 56. Kee, R. J.; Rupley, F. M.; Miller, J. A. Chemkin-II: A Fortran Chemical Kinetics Package for the Analysis of Gas-Phase Chemical Kinetics (Sandia National Laboratories Report No. SAND89-8009B, 1991). 57. Frenklach, M.; Wang, H.; Yu, C.-L.; Goldenberg, M.; Bowman, C. T.; Hanson, R. K.; Davidson, D. F.; Chang, E. J.; Smith, G. P.; Golden, D. M.; Gardiner, W. C.; Lissianski, V. GRI-Mech 1.2, http://www.gri.org. 2 9 8

0

In Computational Thermochemistry; Irikura, K., et al.; ACS Symposium Series; American Chemical Society: Washington, DC, 1905.